note larsen [andye] is right. Fitness function <i>f</i> must be continuus, so that for two similar individuals <i>x₁</i> and <i>x₂</i>, <i>f(x₁)</i> is similar to <i>f(x₂)</i>. <p>So, <strong>small causes (small random mutations) imply small changes (small variations of fitness)</strong>.<p> If you want to apply the ideas of GA to evolve programs, you should explore the so-called field of [http://www.cs.ucl.ac.uk/research/genprog/gp2faq/gp2faq2.html|Genetic Programming]; a classic reference are Holland's [http://www.amazon.com/exec/obidos/search-handle-url/index=books&field-author=Holland%2C%20John%20H./104-5341406-3579968|books] 82449 82502